Polarized localization of kinesin-1 and RIC-7 drives axonal mitochondria anterograde transport

Mitochondria are complex organelles with multiple functions including ATP production, calcium buffering, reactive oxygen species signaling, and programmed cell death. In neurons, synaptic mitochondria support neurotransmission by at least two mechanisms: serving as an energy source via ATP production, and acting as a reservoir for calcium buffering (Ivannikov et al., 2013; Kang et al., 2008; Kwon et al., 2016; Sun et al., 2013; Tang and Zucker, 1997; Verstreken et al., 2005). Therefore, maintaining mitochondria distribution is critical for neuronal homeostasis. Indeed, mitochondria mislocalization can lead to neurodegeneration, and defects in mitochondria trafficking are found in a variety of disease models for Alzheimer’s disease, Parkinson’s disease, and amyotrophic lateral sclerosis (Baldwin et al., 2016; De Vos et al., 2007; Ding et al., 2022; López-Doménech et al., 2016; Nguyen et al., 2014; Rawson et al., 2014; Rui et al., 2006; Wang et al., 2010, 2011; Zhao et al., 2010).

Although evidence suggests that mitochondria biogenesis can occur in the axon, it is believed that most mitochondria are generated in the cell body and transported to distal processes (Amiri and Hollenbeck, 2008; Davis and Clayton, 1996). Mitochondria transport is primarily driven by the microtubule plus-end–directed motor kinesin-1 and the minus-end–directed motor dynein (Hurd and Saxton, 1996; Pilling et al., 2006). Microtubules in the axon are predominantly oriented with their plus end pointing toward the distal tip (plus-end-out). Therefore, kinesin-1 mainly drives the exit of mitochondria from the cell body into the axon and anterograde trafficking toward the distal end, whereas dynein moves mitochondria retrogradely in the axon toward the cell body.

The kinesin-1 and dynein motors bind mitochondria through adaptor complexes, such as Miro and Milton/TRAK (Amiri and Hollenbeck, 2008; Glater et al., 2006; Guo et al., 2005; Stowers et al., 2002; van Spronsen et al., 2013). Miro is a conserved GTPase with a C-terminal transmembrane domain that inserts into the outer mitochondria membrane (Fransson et al., 2003). Miro binds to Milton, a coiled-coil protein that in turn binds to kinesin-1 and to dynein (Glater et al., 2006; van Spronsen et al., 2013). In vitro, Miro can also bind to KIF5/kinesin-1 directly (Macaskill et al., 2009). In Drosophila photoreceptors and motor neurons, loss-of-function of either Miro or Milton leads to dramatic loss of mitochondria in axons and depletion of mitochondria at the neuromuscular junctions, suggesting that Miro and Milton are essential for mitochondria exit from the cell body and axonal anterograde traffic (Guo et al., 2005; Stowers et al., 2002). In mice, Miro1, as opposed to Miro2, is the major regulator of mitochondria traffic. Interestingly, cultured hippocampal neurons from Miro1 knockout mice still show normal amounts of mitochondria in the axon, although fast axonal mitochondrial transport is dramatically decreased (López-Doménech et al., 2016), suggesting that there are Miro1-independent mechanisms to move mitochondria into the axon in mice, potentially via functional compensation from Miro2, or another unknown adaptor. In addition, a recent study suggests that Miro proteins are not required for TRAK/kinesin-dependent anterograde movement in mouse embryonic fibroblasts (López-Doménech et al., 2018). These conflicting data suggest that our understanding of mitochondria transport in vivo is incomplete.

Studies in Caenorhabditis elegans also support the existence of Miro/TRAK-independent mechanisms of mitochondria transport. First, TRAK is not essential for kinesin-1-based transport (Zhao et al., 2021). Second, Miro is not required for axonal mitochondria distribution in some unipolar neurons such as touch receptor neurons and PVQ neurons (Ding et al., 2022; Sure et al., 2018). Third, a recent genetic screen identified the metaxins MTX-1 and MTX-2, components of the mitochondrial sorting and assembly machinery, as important regulators for mitochondria transport (Zhao et al., 2021). In the DA9 neuron, genetic analysis indicated that mtx-1/2 and miro-1 redundantly regulate kinesin-1-mediated mitochondria transport and that overexpression of mtx-1 or mtx-2 in the PVD neuron can substitute for MIRO-1’s function in kinesin-1 based transport but not vice versa (Zhao et al., 2021).

Another protein, RIC-7, is essential for mitochondria localization in C. elegans. In ric-7 mutants, mitochondria are absent from the axons, similar to kinesin-1 mutants (Rawson et al., 2014). RIC-7 does not contain known conserved domains or motifs and does not have an obvious homolog in vertebrates (Rawson et al., 2014). RIC-7 is localized on mitochondria, but how it regulates mitochondria localization in neurons is not understood (Rawson et al., 2014).

Here, we analyze mitochondria transport in vivo using the C. elegans AIY and DA9 neurons as model systems. We interrogate the mitochondria transport machinery by tagging the endogenous proteins and visualizing their localization and dynamics in physiological conditions, as well as by genetic analysis using deletion mutants. We find that Miro is important for robust mitochondria anterograde trafficking in axons but is not absolutely required, whereas it is essential for retrograde axonal trafficking. We demonstrate that RIC-7 is specifically required for kinesin-1-based mitochondria trafficking by localizing with kinesin-1 to the leading end of mitochondria. Further, Miro/MIRO-1 and metaxin2/MTX-2 function together to regulate the localization of RIC-7. Using structure/function analysis, we identify the domains in the RIC-7 protein that are crucial for its localization and function. Our work provides new insights into the cellular mechanisms that govern mitochondria traffic in living nematodes and identifies polarized protein localization at the leading end as a novel functional site critical for mitochondria movement.

To assess mitochondria distribution in vivo, we use a matrix-targeted EGFP marker that shows consistent morphology of mitochondria in AIY neurons (see Materials and methods). The AIY neurons are a pair of unipolar interneurons that each extends a single neurite anteriorly, with microtubule plus ends orienting toward the neurite tip (Hill et al., 2019), and have been well established as a model system to study presynaptic assembly and synaptic vesicle localization (Colón-Ramos et al., 2007; Stavoe and Colón-Ramos, 2012; Stavoe et al., 2012; White et al., 1986).

In AIY, partial loss-of-function of unc-116, the sole ortholog of C. elegans kinesin-1, results in dramatic loss of mitochondria in the neurite (Fig. 1), consistent with an essential role for kinesin-1 in axonal mitochondria anterograde transport (Hurd and Saxton, 1996; Pilling et al., 2006; Rawson et al., 2014). To assess the role of TRAK/Milton and Miro, we used CRISPR to generate deletion alleles of trak-1, the sole C. elegans TRAK ortholog, and of miro-1, -2, and -3, the three C. elegans Miro orthologs. We found that mitochondria localization is normal in AIY neurites in trak-1(KO) mutants and in a previously published trak-1 loss-of-function allele (Zhao et al., 2021) (Fig. 1), consistent with the finding that TRAK-1 is not required for axonal mitochondria anterograde traffic (Zhao et al., 2021). Surprisingly, we found significantly more mitochondria in the AIY neurites of miro-1(KO) animals. This is not due to compensation from the other two miro genes, as a similar distribution was observed in the miro-1/2/3 triple knockouts (Fig. 1). The similar phenotypes between the miro-1 single mutant and the triple mutant indicate that miro-1 is the major functional ortholog. Further, the different phenotypes of miro and trak mutants indicate that Miro and TRAK do not have equivalent functions in mitochondria transport, as recent studies have suggested (López-Doménech et al., 2018; Zhao et al., 2021). Overall, these results suggest the existence of Miro- and TRAK-independent mechanisms that are dependent on kinesin-1.

To exclude the possibility that excessive axonal mitochondria in miro-1 mutants arise from local biogenesis, we examined overall mitochondria mass during development. We focused on the miro-1 single mutant since it showed similar mitochondria distribution as miro-1/2/3 mutants. We used the total fluorescence of our mitochondria marker to estimate relative mitochondria mass and normalized this measurement to that of the control larval stage 1 (L1) animals. Using this assay, we found that there is no significant difference in mitochondria mass between control and miro-1 mutants at any developmental stage. In both control and miro-1 mutants, total mitochondria mass increases by ∼40% from L1 to the larval stage 2 (L2) but does not change significantly from L2 to L4 (Fig. 2 A). Next, we examined mitochondria mass specifically in neurites. In control, neurite mass increased dramatically from L1 to L2 (0.12 versus 0.25, P = 0.009) but did not change significantly from L2 to L4 (0.25 versus 0.31, P = 0.79). Strikingly, miro-1 mutants had a similar amount of neurite mitochondria as L1s but accumulated slightly more as L2s (L1: 0.1, L2: 0.37, P < 0.0001) and dramatically more as L4 (0.77, P < 0.0001 compared with L2) (Fig. 2 B). The increase in neurite mitochondria from L2 to L4 in miro-1 mutants was accompanied by a corresponding loss of mitochondria in the cell body (Fig. 2 C), consistent with the overall stability of mitochondria total mass (Fig. 2 A). Overall, these results suggest that there is an increased net flux of mitochondria in miro-1 mutants from the cell body into the AIY neurite—in the anterograde direction.

To confirm and extend this result, we used a photoswitchable fluorophore to perform a “pulse-chase” experiment on mitochondria traffic, enabling assessment of mitochondria transport at much longer time scales than is possible by continuous live imaging. This assay is powerful in that it integrates rare or short traffic events over a long period of time. We expressed mitochondria matrix-targeted photoconvertible fluorescent protein Dendra2 and photoconverted mitochondria either in the cell body or in the neurite and imaged after a certain period. Compared with the control animals, mitochondria photoconverted in the cell body showed retarded anterograde traffic to the neurite in miro-1 mutants. However, after 8 h, the amount of mitochondria that move to the neurite is equivalent to that in controls at 4 h, suggesting that MIRO-1 independent mechanisms are sufficient to move mitochondria into the neurite, albeit less efficiently (Fig. 2, D and F). Consistently, the average gross speed of anterograde trafficking in miro-1 mutants is reduced by >50%, from 2.9 μm/min to 1.1 μm/s on average (Fig. 2 G; and Videos 1 and 2). By contrast, mitochondria photoconverted in the neurite were retained even after 24 h in miro-1 mutants, while 80% were moved back to the cell body in control animals (Fig. 2, E and H). These results indicate that MIRO-1 is essential for mitochondria retrograde movement. Interestingly, although in some cases, Milton/TRAK acts as the adaptor between Miro and dynein (van Spronsen et al., 2013), we found that mitochondria photoconverted in the neurite in trak-1 mutants were transported back to the cell body, similar to controls, suggesting that TRAK-1 is not required for retrograde movement in AIY unlike Miro (Fig. 2, E and H). Overall, the photoconversion data are consistent with the mitochondria mass experiments and indicate that mitochondria traffic is anterogradely biased in AIY in miro mutants, resulting in axonal accumulation and cell body depletion over time.

Together, our results clarify the function of Miro in vivo in C. elegans. Miro is important for robust mitochondria movements, both anterograde and retrograde. However, in the absence of Miro, mitochondria transport persists in the anterograde direction, while retrograde transport is completely blocked.

Unlike miro mutants, but similar to kinesin-1 mutants (Fig. 1), ric-7 mutants lack mitochondria in neuronal processes (Rawson et al., 2014). Consistently, we found that mitochondria are stuck in the AIY cell body in ric-7 mutants (Fig. 3, A and D). We also performed electron microscopy on wild type and ric-7 mutants and examined the neuropil, which consists of ∼180 axons (Fig. S1 A, “neuropil”). Consistent with fluorescent marker data, we found that in ric-7 mutants mitochondria are absent from the entire neuropil. In the soma, mitochondria accumulated in ric-7 animals, but size and morphology were largely unaffected (Fig. S1, A and B), suggesting that ric-7 is specifically required for mitochondria localization.

Previous studies did not determine whether ric-7 mutants lack mitochondria specifically in the axon or also in the dendrite. As microtubule polarity is different in these compartments, and thus the underlying trafficking mechanisms could vary, we sought to test if ric-7 is required for general mitochondria localization, or if its function is specific to axonal mitochondria. Therefore, we investigated ric-7’s role in mitochondria transport in the DA9 neuron, a bipolar excitatory motor neuron with a dendrite in the ventral nerve cord and an axon crossing the body and joining the dorsal nerve cord, where it synapses on the body wall muscle (Fig. 3 C). In DA9, microtubules are plus-end-out in the axon and predominantly minus-end-out in the dendrite (Yan et al., 2013), and axonal mitochondria anterograde traffic has been shown to depend entirely on kinesin-1, but only partially on MIRO-1 (Zhao et al., 2021).

We found that mitochondria are absent from the DA9 axon in ric-7 mutants but are still present in the dendrite, indicating that ric-7 is specifically required to transport mitochondria into axons (Fig. 3, B, E, and F). Consistently, both anterograde and retrograde traffic in the dendrite are not affected in ric-7 mutants (Fig. S2 A), although the dendritic mitochondria are significantly longer than WT (Fig. S2 B).

We previously observed mitochondria anterograde transport in miro mutants in the AIY neurite (Fig. 2 D). Therefore, we tested whether this Miro-independent form of mitochondria traffic is ric-7 dependent. We found that regardless of the presence or absence of Miro, mitochondria are absent from the AIY neurite in ric-7 mutants (Fig. 3 D). Thus, anterograde traffic in AIY neurite in the absence of MIRO-1 is also RIC-7 dependent. The AIY neurite and the DA9 axon are similar in that both have plus end-out microtubules, while the DA9 dendrite has predominantly minus end-out microtubules. Together, these results indicate that RIC-7 is specifically essential for the plus end–directed movement of mitochondria—that is, the movement that is mediated by the kinesin-1 motor.

To better understand how RIC-7 regulates mitochondria transport, we used the native and tissue-specific fluorescence (NATF) approach (He et al., 2019) to visualize the endogenous localization of RIC-7 in the DA9 neuron. Briefly, 7xGFP11 was inserted at the endogenous C-terminus of RIC-7 by CRISPR and the complementary GFP1-10 fragment was overexpressed in DA9 (Fig. 4 A). This tagging approach does not affect RIC-7’s function as mitochondria localization is not altered (Fig. S3 A). Hereafter, we refer to the endogenous RIC-7 molecules labeled by reconstituted split-GFP as RIC-7::7xSplitGFP. We found that RIC-7::7xSplitGFP formed puncta that are all colocalized with mitochondria (Fig. 4, B–D), although we cannot exclude that some RIC-7 is independent of mitochondria below our detection threshold. By contrast, not every mitochondrion associates with visible RIC-7::7xSplitGFP puncta. We found that the percentage of mitochondria associated with RIC-7::7xSplitGFP is significantly higher in the axon than in the dendrite (80% in axons; 45% in dendrites; P < 0.0001) (Fig. 4 E). Thus, like overexpressed RIC-7 (Rawson et al., 2014), endogenous RIC-7 is highly localized to mitochondria, particularly in the axons.

Since RIC-7 is required for mitochondria localization specifically in the axons (Fig. 3), we speculated that it might be associated with mitochondria that are moving in the anterograde direction in the axon. To test this idea, we performed time-lapse imaging and examined the association of RIC-7 with mobile mitochondria in axons and dendrites. Consistent with previous findings that mitochondria motility is less dynamic in vivo and in mature neurons (Lewis et al., 2016; Smit-Rigter et al., 2016), mitochondria trafficking events are rare in DA9—only ∼13% of mitochondria are mobile in the axon during the window of imaging (10 min per video, 10% for anterograde, 3.3% for retrograde, Fig. S3 B). We observed that anterograde mitochondria in the axon (plus-end–directed) invariably (100%, 32/32) associate with RIC-7::7xSplitGFP puncta in contrast to 23% (3/13) of retrograde mitochondria associated with RIC-7 (Fig. 4 G and Video 4). This localization bias is eliminated in the dendrite (Fig. 4 H). The strict requirement for RIC-7 in mitochondria axonal localization and the strong association of RIC-7 with mitochondria undergoing axonal anterograde movement together suggest that RIC-7 acts on mitochondria to enable their anterograde traffic in the axon.

Interestingly, we observed that the precise localization of RIC-7 on individual mitochondria is sometimes polarized—RIC-7 appears highly localized to one end (Fig. 4, B and C). We found that 28% of all axonal mitochondria have RIC-7 puncta at their distal end, closer to the axon tip. By contrast, only 3% of all axonal mitochondria have RIC-7 puncta at their proximal end, closer to the cell body. This polarized localization pattern is reversed in the dendrite: 0% of all dendritic mitochondria have RIC-7 puncta at their distal end, closer to the dendrite tip, and 25% have RIC-7 puncta at their proximal end (Fig. 4 F). Thus, RIC-7 tends to localize to the mitochondria end that correlates with the direction of the plus end of local microtubules—that is, the leading end for kinesin-1-mediated transport.

We speculated that mitochondria anterograde traffic might be correlated with the polarized localization of RIC-7 at the leading end. We tested this idea using time-lapse imaging. We observed that anterograde trafficking mitochondria in the axon often have RIC-7::7xSplitGFP puncta at their leading end (Fig. 4, I and J; and Videos 3 and 4). However, although we could tell that all anterogradely moving mitochondria are associated with RIC-7 (Fig. 4 G), some mitochondria are too small to allow us to resolve the distribution of RIC-7 on mitochondria. Therefore, to better quantify the relationship between RIC-7’s polarized localization and mitochondria traffic, we examined mitochondria exiting the cell body. As previously shown, mitochondria exiting the cell body usually first extend into the neurite, followed by fission and trafficking (Zhao et al., 2021) (Fig. 4 G). Strikingly, as mitochondria extend into the axon, we found that RIC-7 accumulates at the mitochondria tip, leading its way into the axon (Fig. 4, K and L; and Video 5). This happened to all mitochondria we observed entering the axon (8/8, Fig. 4 L). Importantly, we did not observe RIC-7 accumulation at the leading end of mitochondria entering the dendrite (Fig. 4 L and Video 6), indicating that the polarized distribution of RIC-7 is not a pure effect of movement. Overall, these data suggest that RIC-7 acts at the leading end of mitochondria to allow kinesin-1-mediated movement.

As RIC-7 localization is correlated with the directionality of kinesin-1 movement, we labeled endogenous kinesin-1/UNC-116 using the NATF approach (He et al., 2019) and visualized its distribution along with mitochondria in DA9 neurons (Fig. 5, A and B). Although kinesin-1 is known to transport mitochondria, how it is distributed on motile mitochondria has not been reported previously. We found that insertion of 3xGFP11 at the C-terminus doesn’t interfere with the function of kinesin-1, as these animals show normal body locomotion, mitochondria localization, and trafficking (Fig. 5, A and B; and Video 7). (We note that 7xGFP11 did interfere with kinesin-1/UNC-116 function; we hypothesize that the larger 7× tag may interfere with important interactions.) Endogenous kinesin-1::3xSplitGFP is diffused in the cell body, and throughout the processes, similar to the localization of the endogenous kinesin-1 in Drosophila (Kelliher et al., 2018). This diffuse distribution is also consistent with the idea that most kinesin-1 molecules are in the autoinhibited conformation (Blasius et al., 2007). In addition to this diffuse pool, kinesin-1::3xSplitGFP also forms visible puncta in the cell body and in the axon (Fig. 5 B). Overall, kinesin-1::3xSplitGFP is nearly undetectable in the dendrite, consistent with its plus-end–directed motor activity (Fig. 5, B and C). Live imaging shows that kinesin-1::3xSplitGFP undergoes extensive anterograde and retrograde trafficking in the axon, with no significant directional bias (Fig. 5, D and E). The average speed of anterograde trafficking is 0.97 μm/s, consistent with the reported kinesin-1 velocity measured in vitro (Schnitzer et al., 2000; Verbrugge et al., 2009) (Fig. 5 F). The average speed of retrograde trafficking is 1.78 μm/s. These retrograde events presumably reflect the binding of kinesin-1 to cargoes that are transported back to the cell body by dynein, as it was previously demonstrated that kinesins and dynein can bind to the same cargo in axons (Encalada et al., 2011; Hendricks et al., 2010; Szpankowski et al., 2012). The observed average run lengths of anterograde and retrograde trafficking are 5.2 and 7.8 μm, respectively (Fig. 5 G).

Next, we examined the colocalization of kinesin-1::3xSplitGFP with moving mitochondria. Unlike RIC-7, a substantial fraction of axonal kinesin-1/UNC-116 is not associated with mitochondria. Nevertheless, 80% of axonal anterograde mitochondria (8/10) have visible kinesin-1::3xSplitGFP puncta associated with them. The remaining mitochondria were surrounded by diffuse kinesin-1::3xSplitGFP. Similar to the localization of RIC-7, kinesin-1::3xSplitGFP is enriched at the leading end of anterogradely moving mitochondria, albeit to a lesser extent (Fig. 5, H and I; and Video 7). We suspect that the abundant cytosolic pool of kinesin-1 reduces our ability to assess its fine localization on mitochondria, possibly accounting for the lesser degree of polarized localization. Overall, these experiments provide the first view of kinesin-1’s localization on motile mitochondria in vivo and demonstrate that kinesin-1/UNC-116 has polarized localization to the leading end of anterogradely trafficking mitochondria, similar to RIC-7.

Given that kinesin-1 and RIC-7 are both enriched at the leading end of anterograde mitochondria, we sought to examine whether the polarized localization of RIC-7 is dependent on kinesin-1/UNC-116. Mitochondria transport into the axon is blocked in kinesin-1 mutants, making it impossible to examine RIC-7 localization on axonal mitochondria in this background. Therefore, we performed acute cell-specific knockdown of kinesin-1 using the auxin degron system (Fig. 6 A) (Glomb et al., 2023). After 5 h of auxin treatment, mitochondria in the dendrite accumulate at the distal tip, indicating that kinesin-1 is degraded and dynein-mediated transport predominates (Fig. 6 B). By contrast, the number of synaptic mitochondria in the axon stayed constant (Fig. 6, B and D). This suggests that in vivo, synaptic mitochondria are anchored by certain mechanisms that allow them to resist transport imbalances, as previously suggested in vitro (Gutnick et al., 2019). We used these synaptic mitochondria to assess the localization of RIC-7 in neurons with reduced kinesin-1. We found that the polarized localization of RIC-7 is nearly abolished after kinesin-1 knockdown (control: 27.3%; auxin: 4.5%; P < 0.0001) (Fig. 6, B, C, and E). We argue that the loss of polarized localization of RIC-7 is not simply a secondary effect caused by the absence of mitochondria anterograde traffic, as the percentage of mobile mitochondria in normal conditions (10% for anterograde, Fig. S3 B) is far less than the percentage of mitochondria with polarized RIC-7 (27–28%, Fig. 4 F and Fig. 6 E). Rather, this result indicates that kinesin-1 is required for polarized RIC-7 localization on mitochondria, likely mediated via its force generation capacity. Further, kinesin-1 knockdown increases the percentage of mitochondria without apparent colocalization of RIC-7 (control: 35.2%; auxin: 57.1%; P = 0.0006) (Fig. 6 E). The decrease in the overall association of mitochondria with RIC-7 might be due to a difference in detection—unpolarized RIC-7 might simply be more difficult to detect. Alternatively, kinesin-1 may help stabilize the localization of RIC-7 on mitochondria. Taken together, these data indicate that kinesin-1 is required for polarized RIC-7 localization on the mitochondria and that RIC-7 binding to mitochondria is largely independent of kinesin-1.

In summary, we find both RIC-7 and kinesin-1 accumulate at the leading end of axonal anterograde mitochondria, and kinesin-1/UNC-116 controls the localization of RIC-7, supporting the hypothesis that kinesin-1 and RIC-7 work together to drive mitochondria movement toward the microtubule plus end.

Along with MIRO-1, MTX-2 has been identified as an important regulator of mitochondria transport. It is proposed that MTX-2 and MIRO-1 form the adaptor core for both kinesin-1 and dynein motors, with MTX-2 playing the major role (Zhao et al., 2021). Similar to the findings in DA9 and PVD neurons shown in Zhao et al. (2021), the AIY neurite has a reduced number of mitochondria in mtx-2 mutants and a complete loss of mitochondria in mtx-2; miro-1 double mutants (Fig. 7, A and B). We find that RIC-7 is required for mitochondria transport into axons, but not into dendrites (Fig. 3), in contrast to the Miro-1/metaxin2 combination, which is required for transport into both axons and dendrites (Zhao et al., 2021). Like ric-7, EM analysis of miro-1; mtx-2 double mutants shows mitochondria accumulation in the soma (Fig. S1).

To test whether Miro and metaxin2 function independently or together with RIC-7 for kinesin-1-mediated mitochondria transport, we characterized RIC-7 localization in miro-1 and mtx-2 mutants. In the DA9 cell body of miro-1 mutants, RIC-7 becomes much more cytosolic and less enriched in mitochondria (Fig. 7 C). We quantified this distribution in two ways, focusing on mitochondria in the cell body since these are present in all genotypes. First, we quantified the enrichment of RIC-7 on mitochondria by normalizing the intensity of GFP puncta relative to the cytosolic signals (Fig. 7 D, see Materials and methods). Second, we calculated the pixel-by-pixel Pearson’s correlation between RIC-7 and mitochondria: the more evenly RIC-7 is distributed, the higher the correlation will be. Using these approaches, we found that RIC-7 is less enriched in mitochondria in miro-1 mutants (1.8-fold) than in WT (2.8-fold, Fig. 7 D), and is significantly more evenly distributed (less punctate) (Fig. 7 E). Consistently, RIC-7 levels on axonal mitochondria are significantly reduced in miro-1 mutants (Fig. 7 F). Interestingly, RIC-7 is still polarized to the distal end of some axonal mitochondria in miro-1 mutants (Fig. 7, G–I). These results—together with our functional data (Fig. 2)—indicate that MIRO-1 is important for RIC-7’s normal localization and function, but that some of RIC-7’s localization and function remain even in the absence of MIRO-1.

In the cell body of mtx-2 mutants, RIC-7::7xSplitGFP is as enriched on mitochondria as in WT (2.8-fold enrichment). However, unlike in WT where it forms discrete puncta, RIC-7 becomes extremely dispersed over the entire mitochondria, to a greater extent than in miro-1 mutants (Fig. 7, C and E). In the axon, although RIC-7::7xSplitGFP is still localized to mitochondria, the polarized localization at the leading end is significantly reduced, and this is not due to a change in mitochondria length (Fig. 7, G–J; and Video 8). Thus, MTX-2 has a primary role in enabling RIC-7 localization to mitochondria subdomains. We also noticed a significant increase in RIC-7 levels on axonal mitochondria in mtx-2 mutants (Fig. 7 F), raising the possibility that MTX-2 might prevent RIC-7 from binding to mitochondria.

In miro-1; mtx-2 double mutants, mitochondria are trapped in the cell body and fail to traffic into both axon and dendrite (Zhao et al., 2021). Thus, the distribution of RIC-7 on axonal mitochondria cannot be determined. In the cell body, combined effects from single mutants were found: RIC-7 does not form discrete puncta and becomes less enriched (1.3-fold) on mitochondria (Fig. 7, C–E).

Overall, this analysis suggests that MIRO-1 and MTX-2 have different roles in mediating RIC-7’s localization. MIRO-1 is involved in the overall localization of RIC-7 to mitochondria but some RIC-7 can localize to mitochondria even without MIRO-1. By contrast, MTX-2 is required to enrich RIC-7 at specific loci on mitochondria, functioning together with kinesin-1 to polarize RIC-7 distribution toward the plus-end of microtubules.

The previous data suggests that RIC-7 has at least two features. First, to localize to mitochondria, partially independently of MIRO-1. Second, to accumulate at the leading end of mitochondria and enable mitochondria anterograde movement. To better understand how RIC-7 functions, we performed a structure/function analysis. The native genomic sequence of the ric-7 gene spans ∼20 kb and overlaps with two other genes, which would make structure/function analysis at the endogenous locus difficult. Therefore, we decided to perform this analysis using transgenes expressed as single-copy insertions at a standardized genomic locus under the same promoter so that different transgenes will achieve comparable levels of expression. The ric-7 gene encodes for three different splice variants (Fig. S4 B), two of which (ric-7a and ric-7b) can rescue mitochondria localization defects in ric-7 mutants (Rawson et al., 2014). We decided to focus on ric-7b because this isoform is preferentially expressed compared with ric-7a across different neurons (Fig. S4, B and C) (Taylor et al., 2021).

The predicted structure of RIC-7B by Alphafold (Jumper et al., 2021; Varadi et al., 2022) shows three distinct domains: a relatively ordered domain at the N-terminus (1–99 aa), a largely disordered domain in the middle (100–470 aa), and a structured domain at the C-terminus (471–709 aa) that is similar to the OTU domain in human Otulin, according to HHPred and Dali (Holm et al., 2023; Söding et al., 2005) (Fig. 8 A, Fig. S5 A, and Video 9). The N-terminal and the middle regions do not resemble any known structures. Based on this information, we expressed various deletions of RIC-7 to assess their abilities to restore mitochondria localization in ric-7 mutants and to visualize their localization by tagging them at the C-terminus with EGFP.

Expression of full-length RIC-7::EGFP restored abundant axonal mitochondria in ric-7 mutants (Fig. 8 B). Expression of 1–470 aa::EGFP (lacking the region similar to OTU domain) also largely rescued mitochondria localization defects in ric-7 mutants, suggesting that 471–709 aa is mostly dispensable for RIC-7’s function (Fig. 8 B). By contrast, expression of the C-terminal OTU domain 471–709 aa did not restore mitochondria localization in ric-7 mutants (Fig. 8 B). We also used a stronger pan-neuronal promoter Prab-3 to drive the expression of either the OTU domain or the human Otulin or OtulinL, none of which could rescue ric-7 mutants (Fig. S5 B). Consistent with its ability to rescue mitochondria localization in ric-7 mutants, 1–470 aa::EGFP forms puncta that are colocalized with mitochondria in the axon, some of which are at the distal end, similar to endogenously-tagged RIC-7 (Fig. 8, C–E). We further found that expression of either 1–99 aa or 100–470 aa alone cannot rescue mitochondria localization defects in ric-7 mutants, indicating that both domains are required for RIC-7’s function (Fig. 8 B).

Interestingly, RIC-7 1–99 aa::EGFP colocalizes with mitochondria in both WT and ric-7 animals (Fig. 8, C and D; and Fig. S4, D–F). In the axon of WT animals, 1–99 aa::EGFP is colocalized with 99% of mitochondria; however, it is only enriched to the distal end of 10% of mitochondria compared with 21% for 1–470 aa::EGFP, suggesting that 100–470 aa helps RIC-7 form puncta at mitochondria subdomains (Fig. 8 E). We surmise that the weak polarized localization of 1–99 aa may be dependent on endogenous RIC-7, but we cannot test this by examining 1–99 aa localization on axonal mitochondria in the ric-7 mutants as there are no axonal mitochondria. However, consistent with this idea, in the ric-7 mutants, 1–99 aa::EGFP covers the entire mitochondria in the cell body and the dendrite, unlike the endogenous full-length RIC-7, which forms discrete puncta (Fig. S4, D and E; and Fig. 4, C and D). The pan-mitochondria localization of 1–99 aa::EGFP is not dependent on MIRO-1, suggesting that 1–99 aa contains a mitochondria localization signal that either directly binds to mitochondria or mitochondria protein(s) other than MIRO-1 (Fig. S4 D). By comparison, 100–470 aa::EGFP is largely cytoplasmic in WT animals, with only 11% of axonal mitochondria associated with 100–470 aa::EGFP signals. In ric-7 mutant animals, this association is slightly increased (49% of dendritic mitochondria contain EGFP signals), suggesting that 100–470 aa can only weakly bind to mitochondria and is displaced by full-length RIC-7 in WT animals (Fig. 8, C and D; and Fig. S4, D–F).

In summary, we find that RIC-7 1–470 aa is sufficient to support mitochondria transport and that 1–99 aa is required for mitochondria localization, while the disordered domain (100–470 aa) facilitates its polarized localization on mitochondria.

Mitochondria transport is critical to the maintenance of various neuronal functions. In this study, we used the C. elegans AIY neuron and the DA9 neuron as model systems to study mitochondria transport in vivo. We characterized RIC-7 as an essential regulator of kinesin-1-based mitochondria transport in C. elegans. Both kinesin-1 and RIC-7 act at the leading end to drive axonal anterograde mitochondria transport. The N-terminal domain of RIC-7 targets mitochondria, and the disordered domain facilitates RIC-7’s accumulation at the mitochondria leading end. We find that a Miro-independent/RIC-7-dependent mechanism can move mitochondria anterogradely, albeit at a much slower rate. RIC-7 partially relies on MIRO-1 to localize to mitochondria and requires MTX-2 to accumulate at the leading end, where RIC-7 promotes kinesin-1-based mitochondria transport (Fig. 7 H). We propose that the polarized localization of kinesin-1 and RIC-7 underlies anterograde mitochondria trafficking in the axon.

Previous studies in multiple model systems showed that the loss of function of Miro results in variable phenotypes in mitochondria distribution. For example, in Drosophila photoreceptors and motor neurons, the loss of function of Miro leads to a dramatic decrease of mitochondria in axons (Guo et al., 2005; Stowers et al., 2002). By contrast, axonal mitochondria distribution was not altered in hippocampal neurons cultured from Miro1 knockout mice, despite the finding that Miro1 is the primary regulator in fast mitochondria trafficking (Guo et al., 2005; López-Doménech et al., 2016; Stowers et al., 2002). In C. elegans, loss of Miro does not cause a change in mitochondria number in touch receptor neurons and PVQ neurons (Ding et al., 2022; Sure et al., 2018). By contrast, in the C. elegans DA9 neuron, miro loss-of-function leads to reduced mitochondria in the axon (Zhao et al., 2021). This study shows that in the AIY neuron, loss of Miro results in increased mitochondria number. What mechanisms might account for these different phenotypes?

We find that in the absence of Miro, mitochondria anterograde transport is retarded. Further, RIC-7 localization to mitochondria is reduced. One possibility is that the right amount of RIC-7 in mitochondria is essential for processive movements, and without MIRO-1, RIC-7 does not reach this optimal level. In this model, all Miro is doing is to localize RIC-7 to mitochondria. Alternatively, Miro itself may promote kinesin-1-mediated mitochondria transport, such as by strengthening the interaction between kinesin-1 and mitochondria. Nevertheless, over long-time scales, mitochondria can still traffic anterogradely in miro mutants, in a process that depends entirely on RIC-7 and kinesin-1. Cell type–specific differences in this novel form of Miro-independent mitochondria anterograde traffic may account for the variable phenotype observed in Miro mutants. Alternatively, variation in other mechanisms that affect overall mitochondria distribution, such as dynein-mediated retrograde traffic, actin-mediated short-range movements and/or local anchoring, mitophagy, hitchhiking, etc., might underlie these cell-specific differences (Ashrafi et al., 2014; Basu et al., 2021; Guo et al., 2018; Gutnick et al., 2019; López-Doménech et al., 2018).

Our data also support the idea that metaxin2 is partially redundant with Miro for anterograde mitochondria traffic. We find that loss of mtx-2 eliminates mitochondria anterograde traffic in Miro mutants in AIY, consistent with previous results in DA9 and PVD (Zhao et al., 2021). Our results also suggest a model in which Miro helps recruit RIC-7 to mitochondria and MTX-2 facilitates the concentration of RIC-7 into puncta on mitochondria. These distinct roles in regulating RIC-7’s localization may provide an answer to why MTX-2 can substitute for the function of MIRO-1 in regulating kinesin-1-based movement, but not vice versa (Zhao et al., 2021)—it could be that the enrichment of RIC-7 at the leading end of mitochondria is functionally more important than the overall amount on mitochondria. Together, our data suggest that RIC-7 is a key point of convergence for anterograde traffic and that MTX-2 and MIRO-1 work to coordinate the RIC-7 function.

We also analyzed Miro’s role in dynein-based transport. Using photoconversion, we found that mitochondria retrograde trafficking in axons is completely blocked in miro-1 mutants. Consistently, Miro loss-of-function also results in mitochondria disappearance from DA9 dendrites and PVD anterior dendrites in C. elegans (Zhao et al., 2021). Further, in hippocampal neurons cultured from Miro1 knockout mice, dendritic mitochondria are dramatically reduced while axonal mitochondria distribution is not altered (López-Doménech et al., 2016). Altogether, these results highlight Miro’s essential role in dynein-based mitochondria trafficking.

Biochemical and functional analysis suggests that Miro is linked to the dynein/dynactin complex via Milton/TRAK (Brickley and Stephenson, 2011; van Spronsen et al., 2013; Zhao et al., 2021). Surprisingly, we found that mitochondria distribution in the AIY neuron is not altered in trak-1 mutants, and that retrograde trafficking appears normal. These phenotypes are different from those in DA9 and PVD neurons (Zhao et al., 2021), suggesting that the requirement of Trak for retrograde trafficking is context-dependent. One possibility is that another protein is functionally redundant with Trak to link Miro to the dynein complex in AIY. Alternatively, the C. elegans Miro may directly bind to the dynein complex. Indeed, a previous study has shown that human Miro1 can directly bind to kinesin-1 in vitro, without the help of Trak (Macaskill et al., 2009), raising the possibility that the Miro protein may have the capacity to directly bind to motors. These possibilities await further investigation.

Taken together, our data demonstrate that anterograde mitochondria traffic in C. elegans relies on two essential factors, kinesin-1 and RIC-7, which function together with Miro and metaxin.

In this study, we found that RIC-7 is specifically required for mitochondria anterograde trafficking in the axon, facilitating kinesin-1-driven mitochondria movements. How does RIC-7 regulate mitochondria transport? One insight comes from its localization pattern. RIC-7 is preferentially enriched at the leading end of anterogradely trafficking mitochondria in the axon. By contrast, MIRO-1 and MTX-2 are ubiquitously distributed in mitochondria (Fu et al., 2020; Zhao et al., 2021). We also found that UNC-116, the sole ortholog of kinesin-1 in C. elegans, displays similar mitochondria localization as RIC-7. Previous in vitro studies found that kinesin-1 molecules that are artificially attached to liposomes would dynamically accumulate at the tip of an extending tube. A model was proposed in which lagging kinesin molecules will rapidly move to the growing tip, where they accumulate and exert force on the tube (Leduc et al., 2004). Our data indicate that the dynamic accumulation of kinesin motors on membrane compartments occurs in vivo and that RIC-7 displays similar dynamic localization.

How does RIC-7 facilitate kinesin-1-mediated mitochondria traffic? One model is that RIC-7 is an essential part of the adaptor complex and recruits kinesin-1 to mitochondria. As kinesin-1 accumulates at the leading end due to its motor function, the fluid nature of the mitochondria membrane would allow the adaptor complex to gather at the same spot. Indeed, when linked to kinesin-1 by a light-inducible dimerization system, Miro accumulates at the tip of deformed mitochondria (Song et al., 2022). In this model, the clustering of RIC-7 is a passive consequence of kinesin-1 accumulation at the mitochondria leading end. Consistent with this model, we found that kinesin-1 degradation leads to the abolishment of the polarized localization of RIC-7 on mitochondria.

However, RIC-7 may also have a specialized function at mitochondria tips. For example, RIC-7 may help stabilize or organize the clustering of kinesin-1/UNC-116. Theoretical analysis has shown that collective force generated by accumulated kinesin-1 motors at the tip allows tube formation (Leduc et al., 2004). In the context of organelle trafficking, the accumulation of motors greatly facilitates organelle movements. For example, an in vivo study has demonstrated that clustering of dynein into microdomains on the membrane of a phagosome facilitates rapid directed transport of the phagosome toward microtubule minus ends. This clustering was proposed to allow cooperative force on a single microtubule (Rai et al., 2016). In vitro, the clustering of small numbers of kinesin-1 by DNA scaffold enables long-distance transport that normally would require high numbers of individual motors (Jiang, 2021). Given the special localization and cellular function of RIC-7, RIC-7 may act as a scaffold to cluster small numbers of kinesin-1 molecules to enable their fast rebinding to microtubules and promote their processivity. Our finding that RIC-7’s function requires its disordered sequence may hint at this clustering role, as intrinsically disordered regions are known to contribute to scaffolding function and promote the clustering of other proteins (Cortese et al., 2008).

Previous analysis found no obvious mammalian homologs based on the similarity of the primary sequence, and even among nematodes, RIC-7 is highly divergent (Rawson et al., 2014). Using new tools for homology detection (HHPred and Dali), we found that the C-terminus of RIC-7b (471–709 aa) is similar to the human Otulin domain. However, our structure/function analysis showed that this domain is dispensable for RIC-7’s function in mitochondria traffic (Fig. 8 B). By contrast, we found that the 1–470 aa fragment of RIC-7 is necessary and sufficient to rescue mitochondria localization defects in ric-7 mutants. This fragment contains a relatively ordered region (1–99 aa) that can localize to mitochondria and a disordered region (100–470 aa) that is required for RIC-7’s function (Fig. 8). The exact molecular function of this disordered region is unclear. Interestingly, unlike 1–470 aa that mimic RIC-7’s full-length localization, 1–99 aa are not confined to subdomains of mitochondria (Fig. S4, D and E). This resembles the localization of RIC-7 in mtx-2 mutants and in kinesin-1 knockdown, suggesting that 100–470 aa may be involved in the interaction between RIC-7 and MTX-2 or kinesin-1. We speculate that potential functional equivalents of RIC-7 in mammalian systems may also be largely composed of a rapidly evolved disordered region, possibly explaining why this critical factor for mitochondria traffic has evaded homology-based searches.

All strains were maintained at 20°C on NGM plates seeded with OP50 E. coli. Strains used in this study are listed in Table S1.

In initial attempts to label mitochondria in AIY, we tried three different markers: TOMM-20 transmembrane domain, cytochrome c oxidase subunit 4 (COX-4), and the mitochondria matrix localization signal (MLS) from the chicken mitochondrial aspartate aminotransferase as used in pPD96.32 (Fire Lab Vector Kit). Overexpression of TOMM-20::EGFP exhibited inconsistent mitochondria morphology, sometimes tubular and sometimes globular. Overexpression of COX-4::EGFP resulted in mitochondria fragmentation. In contrast, mitochondria labeled by MLS::EGFP (mito::EGFP in this paper) show consistent tubular morphology. Therefore, we decided to pursue this marker. The microinjection mix contained 30 ng/μl Pttx-3::mito::EGFP::let-858 3′UTR, 30 ng/μl DACR18 (Pttx-3::mcherry::rab-3::unc-10 3′UTR), 25 ng/µ Punc-122::GFP, and Promega 1 kb DNA ladder as the filler. The mix was injected into N2, and transgenic animals with integrated extrachromosomal arrays were then generated by the TMP/UV method. Among several integration lines, we selected wpIs127 because it expresses the transgenes at a relatively low level and exhibits consistent mitochondria morphology. wpIs127 was outcrossed five times before further analysis.

Mitochondria in DA9 were labeled by mitochondria matrix localization signal (pPD96.32; Fire Lab Vector Kit) fused with TagRFP and expressed as extrachromosomal arrays. The microinjection mix contained Pitr-1pB::mito::TagRFP::let858 3′UTR at 30 ng/μl, Podr-1::gfp at 20 ng/μl, and Promega 1 kb DNA ladder as the filler.

All plasmids used for expressing transgenes in neurons were assembled to the destination vector pCFJ150 using Gateway recombination (Invitrogen). Gibson reactions were used to generate the entry clones (Gibson et al., 2009). Plasmids that express trak-1 and miro-1 sgRNAs were made by swapping the N₁₉ sequence of the unc-119 sgRNA from the preexisting plasmid (#46169; Addgene) (Friedland et al., 2013) using PCR, followed by 5′ phosphorylation and blunt end ligation.

All plasmids used in this study are listed in Table S2.

trak-1(wp84), miro-1(wp88), miro-2(wp92), and miro-3(wp94) are whole gene deletions generated by NHEJ with CRISPR/Cas9. For trak-1(wp84) and miro-1(wp88), CRISPR/Cas9 was performed using the dpy-10(cn64) as the co-conversion marker, as previously described in Arribere et al. (2014). Briefly, sgRNA and Cas9 were expressed using plasmids. The microinjection mix contained 50 ng/μl sgRNA vector, 40 ng/μl dpy-10 sgRNA vector, 50 ng/μl Cas9 expression vector (pDD162) (Dickinson et al., 2013), and 26 ng/μl dpy-10 repair template AF-ZF-827 (Arribere et al., 2014). The mix was injected into XE2194, and F1 rollers were genotyped by single-worm PCR. miro-2(wp92) and miro-3(wp94) were generated by NHEJ using a preassembled Cas9 ribonucleoprotein complex as described in (Dokshin et al., 2018). The injection mix contains 0.25 μg/μl Streptococcus pyogenes Cas9 (#1081058; IDT), 0.1 μg/μl tracrRNA (#1072532; IDT), 0.056 μg/μl crRNA in total (0.028 μg/μl for each sequence, synthesized by IDT), and 40 ng/μl PRF4 (rol-6 [su1006]) plasmid. miro-2 and miro-3 were simultaneously targeted by the same sgRNAs, and the mix was injected into miro-1(wp88) to generate miro triple mutant. F1 progenies (rollers and non-rollers) were screened by single worm PCR.

ric-7::7xgfp11 knock-in was generated by CRISPR/Cas9 using a preassembled Cas9 ribonucleoprotein complex as previously described (Ghanta and Mello, 2020). Briefly, the injection mix contained 0.25 μg/μl S. pyogenes Cas9 (#1081058; IDT), 0.1 μg/μl tracrRNA (#1072532; IDT), 0.056 μg/μl crRNA (synthesized by IDT), 25 ng/μl premelted double-stranded DNA repair template or 0.11 μg/μl of ssDNA repair template and 40 ng/μl PRF4 (rol-6 [su1006]) plasmid. Non-roller F1 progeny from P0 roller jackpot plates were screened by single worm PCR. Homozygotes from the F2 generation were verified by Sanger sequencing.

For unc-116::3xgfp11 and unc-116::AID knock-in, instead of using tracrRNA and crRNA, a single gRNA of 3.7 μg was used (Glomb et al., 2023). This gRNA was synthesized from ssDNA using the EnGen sgRNA Synthesis Kit, S. pygenes (#E3322S; NEB), and subsequently purified using a Monarch RNA cleanup kit (#T2040; NEB).

All repair templates contained ∼35 bp homology arms on both ends. For tandem gfp11 insertions, a GSGGGG liker was added between the gene of interest and gfp11. For unc-116, we chose to use 3× tandem repeats instead of 7×, as mitochondria trafficking was dramatically reduced when unc-116 was tagged with 7xgfp11. The repair templates for ric-7::7xgfp11 and unc-116::3xgfp11 were synthesized as a Gblock by IDT, amplified by PCR, gel purified, and further cleaned up with the Qiagen MinElute kit. The repair template for unc-116::AID was synthesized by IDT as ssDNA.

ric-7::7xgfp11 repair template (homology arms, linker, 7xgfp11):

5′-GTC​CGG​TTC​CCC​TGT​TCT​ACA​TAC​CGG​ACG​ATG​AAG​GAT​CCG​GAG​GTG​GCG​GGC​GTG​ACC​ACA​TGG​TCC​TTC​ATG​AGT​ATG​TAA​ATG​CTG​CTG​GGA​TTA​CAG​GTG​GCT​CTG​GAG​GTA​GAG​ATC​ATA​TGG​TTC​TCC​ACG​AAT​ACG​TTA​ACG​CCG​CAG​GCA​TCA​CTG​GCG​GTA​GTG​GAG​GAC​GCG​ACC​ATA​TGG​TAC​TAC​ATG​AAT​ATG​TCA​ATG​CAG​CCG​GAA​TAA​CCG​GAG​GGT​CCG​GAG​GCC​GGG​ATC​ACA​TGG​TGC​TGC​ATG​AGT​ATG​TGA​ACG​CGG​CGG​GTA​TAA​CTG​GTG​GGT​CGG​GCG​GAC​GAG​ACC​ATA​TGG​TGC​TTC​ACG​AAT​ACG​TAA​ACG​CAG​CTG​GCA​TTA​CTG​GCG​GAT​CAG​GTG​GCA​GGG​ATC​ACA​TGG​TAC​TCC​ATG​AGT​ACG​TGA​ACG​CTG​CTG​GAA​TCA​CAG​GCG​GTA​GCG​GCG​GTC​GGG​ACC​ATA​TGG​TCC​TGC​ACG​AAT​ATG​TCA​ATG​CTG​CCG​GTA​TCA​CCT​AAT​AAC​TAG​GAC​CCT​CTT​TTT​GTT​ATT​CGT​AAC​TTG​TAA-3′

unc-116::3xgfp11 repair template (homology arms, linker, 3xgfp11):

5′-GAC​AAG​TGT​ATA​CGT​CTC​CGT​CAG​CAG​GAA​TGT​CAG​GAT​CCG​GAG​GTG​GCG​GGC​GTG​ACC​ACA​TGG​TCC​TTC​ATG​AGT​ATG​TAA​ATG​CTG​CTG​GGA​TTA​CAG​GTG​GCT​CTG​GAG​GTA​GAG​ATC​ATA​TGG​TTC​TCC​ACG​AAT​ACG​TTA​ACG​CCG​CAG​GCA​TCA​CAG​GCG​GTA​GCG​GCG​GTC​GGG​ACC​ATA​TGG​TCC​TGC​ACG​AAT​ATG​TCA​ATG​CTG​CCG​GTA​TCA​CCC​AAG​GAG​CTC​CAA​ATG​GCT​CAA​ACG​GTG​TGT​TTA​G-3′

unc-116::AID repair template (Glomb et al., 2023, homology arms, AID):

5′-GAC​AAG​TGT​ATA​CGT​CTC​CGT​CAG​CAG​GAA​TGT​CAA​TGC​CTA​AAG​ATC​CAG​CCA​AAC​CTC​CGG​CCA​AGG​CAC​AAG​TTG​TGG​GAT​GGC​CAC​CGG​TGA​GAT​CAT​ACC​GGA​AGA​ACG​TGA​TGG​TTT​CCT​GCC​AAA​AAT​CAA​GCG​GTG​GCC​CGG​AGG​CGG​CGG​CGT​TCG​TGA​AGC​AAG​GAG​CTC​CAA​ATG​GCT​CAA​ACG​GTG​TGT​TTA​G-3′

All gRNA sequences used in this study are listed in Table S3. Primers used for genotyping are listed in Table S4. Allele sequences are listed in Table S5.

Structure/function analysis of RIC-7 was performed by expressing different transgenes as single-copy insertions at the same genomic locus under the same promoter to achieve comparable levels of expression. As the endogenous expression level of RIC-7 in DA9 is extremely low (Fig. S4 A) (Taylor et al., 2021), we initially used a weak promoter, Pitr-1pB, to drive the expression of transgenes in DA9 (Klassen and Shen, 2007; Taylor et al., 2021). Using this approach, mitochondria localization is restored when ric-7b full length is expressed, suggesting that the expression level is high enough for the functional assessment of the truncated proteins. However, it is not high enough for us to reliably examine the localization of the proteins. Therefore, we switched to a stronger DA9 promoter, Pmig-13 (Klassen and Shen, 2007) (Fig. S4 A), to examine protein localization.

CRISPR/Cas9 was used to generate single-copy insertion at the ttTi5605 landing site on Chromosome I in the strain EG8078. The injection mix contained 0.25 μg/μl S. pyogenes Cas9 (#1081058; IDT), 0.1 μg/μl tracrRNA (#1072532; IDT), 0.028 μg/μl crRNA each (synthesized by IDT), the transgene plasmid constructed in pCFJ150 (100 ng/μl), pMA122 (peel-1 negative selection, 10 ng/μl), and pCFJ90 (Pmyo2::mCherry, 1.5 ng/μl). Each injected P0 animal was singled and grown for 7–8 days at room temperature until starvation followed by heat shock at 34°C for 4 h. The next day after heat shock, P0 plates were checked for potential insertions. Successful insertions should survive the heat shock, not express mCherry, and be non-unc. ∼10 non-unc worms from each P0 plate were singled and verified for insertion by genotyping.

To acutely knock down kinesin-1/unc-116 in DA9 (Glomb et al., 2023), young L4 animals were cultured on NGM plates containing 1 mM Auxin (#A10556.14; Thermo Fisher Scientific) for 5 h before imaging. Animals of the same stage cultured on NGM plates lacking auxin served as controls.

To visualize mitochondria distribution in AIY and DA9, L4 animals were immobilized by 8 mM levamisole and mounted on 3% agarose pad. Still images were acquired with an ORCA-Flash4.0 camera (Hamamatsu) on a VT-iSIM system (BioVision) built around a Leica Laser Safe DMi8 Inverted scope with a PL APO 100×/1.47 OIL CORR TIRF objective for AIY or a HC PL APO 63×/1.40 OIL CS2 objective for DA9. Image acquisition was controlled by MetaMorph Advanced Confocal Acquisition Software Package. The AIY neuron that is closer to the coverslip was imaged. Z-stacks were acquired with a step size of 0.2 μm to capture the whole neuron.

To better image the endogenous proteins and mitochondria in DA9, L4 animals were immobilized with 50 mM Muscimol for 2–3 min before mounting, which increases the chance that either the dorsal or the ventral side faces the coverslip. Images were acquired with a HC PL APO 63×/1.40 OIL CS2 objective. Emission filters 525/50 and 605/52 nm were used for 488 and 561 nm illumination, respectively. Images were taken at 400 ms exposure sequentially for 488 and 561 nm at each Z position. Only the dendrite or the axon was imaged for each animal depending on the orientation. Time-lapse imaging was performed for 10 min with 10-s intervals, with three z-steps of 0.2-μm step size.

To generate the kymograph of kinesin-1::3xSplitGFP (UNC-116::3xSplitGFP) trafficking, images were taken continuously on a single plane at 400 ms exposure with the 488-nm laser using the streaming mode.

For Photoconversion in the AIY neuron, images were acquired on a UltraVIEW VoX spinning-disc confocal microscope (PerkinElmer) built around a Nikon Ti-E Eclipse equipped with a 60×/1.4NA CFI Plan Apo objective and a Hamamatsu C9100-50 camera and driven by the Volocity software (Improvision). Mitochondria in the cell body or the axon in both AIY neurons were photoconverted with a 405 nm laser using the PhotoKinesis Unit. After photoconversion, worms were recovered and placed on NGM plates with OP50. After a certain period (4, 8, or 24 h), these worms were imaged again. All animals were staged to be photoconverted so that the endpoint was at L4. For time-lapse imaging, young L4 animals were immobilized by 8 mM levamisole for 5 min and then washed with M9 at least five times before mounting on a 10% agarose pad, which is in turn mounted on a gas permeable slide to minimize hypoxia. Videos were taken with 0.3 μm z-step size at 1–2 min intervals for WT for a total duration of 1–2 h and 5-min intervals for miro-1 mutants for 4–6 h. We were not able to assess the detailed characteristics of mitochondria transport in miro-1 mutants as this requires shortening the imaging interval, and we found that mitochondria trafficking in miro-1 mutants was very sensitive to laser exposure.

All fluorescent images were acquired at room temperature (∼22°C).

To quantify mitochondria in the AIY axon, we used two different metrics: (1) the fraction of the total fluorescence of mito::EGFP in the axon; (2) the number of mitochondria in the axon. We arrived at the same conclusion using both metrics. The first metric is defined by the following formula: the total intensity of the axon/(the total intensity of the cell body + the axon), which was calculated with a custom Matlab script, using maximum projection images. Briefly, images were imported to Matlab using the function “imread.” After background subtraction, images were threshold-segmented using the functions “adaptthresh” and “imbinarize.” The total intensity of the cell body and the axon were then calculated in Matlab by integrating each pixel under the segmented images, respectively. To quantify mitochondria in DA9 axon, only the second metric was used.

To quantify the enrichment of RIC-7::7xSplitGFP and kinesin-1::3xSplitGFP (UNC-116::3xSplitGFP) at the leading end of anterogradely trafficking mitochondria, line scans of the GFP fluorescence intensity along the length of mitochondria were generated in FIJI. After background (intensities outside of the neuron but inside the worm) subtraction, the intensity values were averaged across different timepoints over the course of the movements, including the period when mitochondria were paused, and then normalized to the maximum value of all time points. To average across different mitochondria that differ in length (Fig. 5 I), the values at 51 regularly spaced points along the mitochondria length were estimated using linear interpolation with the interp1 function in Matlab.

Kymographs of kinesin-1::3xSplitGFP (UNC-116::3xSplitGFP) trafficking were generated and analyzed using KymoToolBox.

To compare RIC-7 localization in different genotypes, two different methods were used: (1) enrichment on mitochondria (Fig. 7 D); (2) Pearson’s correlation (Fig. 7 E). Enrichment on mitochondria is defined as the ratio between the intensity of RIC-7::7xSplitGFP puncta on mitochondria and the cytosolic GFP signals, both of which were manually measured in ImageJ and background subtracted. For mtx-2 and miro-1; mtx-2 double mutants where RIC-7::7xSplitGFP does not form discrete puncta, the GFP intensity on the entire mitochondria was measured. For Pearson’s correlation, the DA9 cell body excluding the nucleus was manually specified in ImageJ, and then a customized macro in ImageJ was used to get the intensities of each pixel in both channels. Pearson’s Correlation between RIC-7::7xSplitGFP and mito::TagRFP was calculated using the Matlab function corrcoef.

For ultrastructural analysis of N2, ric-7, miro-1; mtx-2, and unc-116 mutants, we subjected young adult worms to high-pressure freezing (EM ICE, Leica Microsystems). Type-A (3-mm diameter; Technotrade) specimen carriers were immersed in Hexadecane and blotted on a piece of filter paper so that the sides of carriers specimens mounted were coated with a thin layer of Hexadecane. This coating allows easy removal of specimens from carriers after processing. The 100-µm side of a Type-A specimen carrier was first filled with OP50 to the rim of the carrier. Using a worm pick, ∼10 adult animals were placed on top of OP50. A flat side of a Type-B specimen carrier (3-mm diameter; Technotrade) was immersed in Hexadecane and placed on top of the Type-A carrier without blotting such that residual Hexadecane would fill any remaining space between animals and the Type-B carrier. This process would ensure no air bubbles were trapped inside the specimen carrier sandwich. Once assembled, the specimen carrier was loaded into a high-pressure freezer and frozen immediately. Specimens were kept under liquid nitrogen until further processing.

Once all samples were frozen, they were transferred into freeze-substitution media containing 1% osmium tetroxide, 1% glutaraldehyde, 1% water, and anhydrous acetone, which was precooled in an automated freeze substitution unit (AFS 2, Leica Microsystems) to −90°C. The freeze-substitution was carried out with the following protocol: −90°C for 48 h, 5°C/h to −20°C, −20°C for 12 h, and 10°C/h to 20°C. Once completed, specimens were infiltrated into plastic (Epon-Araldite, TedPella, Epon, 6.2 g; Araldite, 4.4 g; DDSA, 12.2 g; and BDMA 0.8 ml) over ∼20 h (30% Epon-Araldite/acetone for 3 h, 70 % for 4 h, 90 % overnight). On the next day, specimens were embedded into 100% Epon-Araldite and cured for 48 h. 70-nm serial sections were prepared using an ultramicrotome (UC7, Leica Microsystem) and collected onto Pioloform (EMS)-coated grids (single slot, Ted Pella). Roughly 200 sections were collected from each specimen around the head ganglia region. After staining with 2.5% uranyl acetate, sections were imaged on a transmission electron microscope (Talos L120C; Thermo Fisher Scientific) equipped with Ceta Camera (Thermo Fisher Scientific). The number of mitochondria and diameter of each mitochondrion were measured blind to genotypes using Fiji. The data are plotted in Prism (GraphPad, v10).

All statistical analysis was performed in GraphPad Prism. Two-sided Fisher’s exact tests were used for binary data. For non-binary data, unless otherwise noted, two-sided t test was used to compare two groups, and one-way ANOVA test followed by Dunn’s multiple comparisons test was used to compare multiple groups. Data distributions were assumed to be normal, but this was not formally tested.

Fig. S1 shows the electron microscopy data of the soma and/or neuropil area in WT, ric-7, miro-1; mtx-2, and kinesin-1 mutants. Fig. S2 shows the quantification of mitochondria trafficking events in the DA9 dendrite in WT and ric-7 mutant. Fig. S3 shows the mitochondria distribution and the percentage of mobile mitochondria in the DA9 axon in the strain where RIC-7 is endogenously tagged with 7xSplitGFP. Fig. S4 shows extended data of the structure/function analysis of RIC-7. Fig. S5 shows that human Otulin or OtulinL or the C-term domain of RIC-7 cannot rescue mitochondria localization in ric-7 mutant. Videos 1 and 2 show mitochondria trafficking in the AIY neuron in WT and miro-1 animals, respectively. Videos 3, 4, 5, and 6 show mitochondria trafficking in different places in the DA9 neuron in the strain where RIC-7 is tagged with 7xSplitGFP (S3: an axon segment close to the cell body; S4: in the synaptic region in the DA9 axon; S5: cell body and the proximal axon; and S6: cell body and the proximal dendrite). Video 7 shows mitochondria trafficking in the DA9 neuron in the strain where kinesin-1 is endogenously tagged with 3xSplitGFP. Video 8 shows mitochondria trafficking in the DA9 neuron in the mtx-2 mutant where RIC-7 is endogenously tagged with 7xSplitGFP. Video 9 shows the 3D structure comparison between RIC-7b and Otulin prepared by Dali. Tables S1, S2, S3, S4, and S5 list the strains, plasmids, gRNA sequences, primers, and allele sequences used in this study, respectively.

All data are available in the main text or the supplementary materials.

We would like to thank members of the Hammarlund Lab and Yogev Lab for technical support and discussion, Dr. Shawn Ferguson for suggestions on the structure/function analysis, Dr. Erik E. Griffin, Dr. Jun Hyun Park, Dr. Lauren Panzera, and Hadas Dabas for providing feedback on the manuscript. Some of the C. elegans strains were provided by the International Caenorhabditis Genetics Center.

This work was supported by National Institutes of Health (grants R01 NS094219 to M. Hammarlund, R35 NS132153 to S. Watanabe, and R01 NS114400 to S. Yogev).

Author contributions: Y. Wu and M. Hammarlund conceived the study. Y. Wu performed and analyzed most of the experiments. C. Ding contributed to the structure/function analysis. B. Sharif and S. Watanabe performed the EM microscopy and analysis. A. Weinreb performed the analysis of ric-7 splicing variants and helped with image analysis. G. Swaim and S. Yogev created the unc-116::AID strain. H. Hao helped with photoconversion experiments and image analysis. The manuscript was written by Y. Wu and M. Hammarlund, with input from all coauthors.